Escolar Documentos
Profissional Documentos
Cultura Documentos
Method 2
78
85
93
66
99
Method 3
97
89
81
73
77
Automobile insurance appraisers examine cars that have been involved in accidental collisions and
estimate the cost of repairs. An insurance executive claims that there are significant differences in the
estimates from different appraisers. To support his claim he takes a random sample of six cars that
have recently been damaged in accidents. Three appraisers then estimate the repair costs of all six
cars. From the data shown below, can we infer at the 5% significance level that the executives claim
is true?
Estimated Repair Cost
Car
Appraiser 1 Appraiser 2 Appraiser 3
1
650
600
750
2
930
910
1010
3
440
450
500
4
750
710
810
5
1190
1050
1250
6
1560
1270
1450
Use SSC = 52877.78, SSR = 1844311.11, SST = 1932644.44
Source of Variation
Treatments
Blocks
Error
Total
H 0 : 1 2 3
SS
52,877.78
1,844,311.11
35,455.56
1,932,644.44
df
2
5
10
17
MS
26,438.889
368,862.222
3545.556
F
7.457
104.035
P-value
0.01042
0.00003
H 0 : 1 2 3
H1 :
At least two means differ
Grand mean = 80.79
SSC = 4(73.25 80.79)2 + 5(84.2 80.79)2 + 5(83.4 80.79)2 =
319.607
SSE = 3(7.46)2 + 4(12.91)2 + 4(9.63)2 = 1204.75
SS
Treatments 319.607
Error
1,204.75
Total
1,524.357
df
2
11
13
MS
159.804
109.523
e12 =
H1 :
At least two means differ
e21 =
F
1.459
e22 =
(36)(95)
380
(36)(95)
380
(344)(95)
380
(344)(95)
380
Quality
Unacceptable
Acceptable
Sub-total
F critical
4.103
3.326
2 =
0.766794742
0.587974176
0.551349659
25139.78873
50
df
4
45
49
SS
40585376295
28440403984
69025780279
MS
1.01E+10
6.32E+08
F
16.05411
Intercept
People
Income
Comptors
Price
Coefficients
-68363.15239
6.439423362
7.272304922
-6709.432007
15968.76477
Standard Error
78524.72511
3.70511702
0.935794757
3818.542618
10219.02625
t Stat
-0.87059
1.737981
7.771261
-1.75707
1.56265
P-value
0.388597
0.089054
7.43E-10
0.085709
0.125141
1.
2.
3.
4.
5.
6.
Equation
Model Useful?
Which independent variable contributes?
Proportion of variation unexplained? Explained?
Standard Error of Estimate:
Multicollearity?
Revenue
1
0.174028
People
1
Income
=9
e14 =
= 86
e23 =
= 86
e24 =
(11 9) 2
9
Price
(8 9) 2
9
(8386)
86
(8786)2
86
2
Comptors
(36)(95)
380
(36)(95)
380
(344)(95)
380
(344)(95)
380
=9
=9
= 86
= 86
Manufacturer
A
B
C
D
Sub-total
12 (9) 8 (9)
5 (9) 11 (9)
36
83 (86) 87 (86) 90 (86) 84 (86)
344
95
95
95
95
380
(8386)
86
Correlation matrix
Revenue
People
Significance F
3.0831E-08
e13 =
(12 9) 2
9
ANOVA
Regression
Residual
Total
=9
(5 9) 2
9
(8386)2
86
2
(9086)
+ 86
Income
Comptors
Price
0.717964
0.085597
0.009324
0.059361
0.153526
-0.09333
1
0.330347
-0.19528
1
-0.15117
(8386)
86
(8486)
+ 86
= 3.68
H0 :
The test score of course 1 is the same as the test score of course 2
H1 :
The test score of course 1 is lower than the score of course 2
Course 1
Rank
Course 2
rank
14
1.5
20
11.5
21
13.5
18
8
17
6.5
22
15
14
1.5
15
3.5
17
6.5
23
16
19
9.5
21
13.5
20
11.5
19
9.5
16
5
15
3.5
W1==55.5
W2==80.5
U1=n1*n2 + n1(n1+1)/2 W1 = 8(8) + 8*9/2 55.5 = 44.5
U2 = n1*n2 + n2(n2+1)- W2 = 8(8) + 8*9/2 80.5 = 19.5
U = 19.5 (the smaller of U1 and U2),
p-value= 0.1172 for U=20 (from table A14)
Decision Rule: reject Ho if p-value< =0.05
Conclusion: Do not reject the null hypothesis.
The marks from course 1 are not lower than those of course 2.
Assume that the test scores are not normally distributed. Test to determine
whether the extra help sessions have been effective at the 5%
significance level.
H0 :
The test score before is the same as the test score after
H1 :
The test score before is lower than the score after
S
t
u
d
e
n
t
A
b
b
y
B
r
e
n
d
a
C
a
r
m
e
n
D
a
v
i
d
E
d
w
a
r
d
F
r
a
n
k
G
i
l
l
d
i
f
f
e
r
e
n
c
e
|
di
ff
er
e
n
c
e|
1
1
1
1
H
e
i
d
i
T 4
Rejection region:
T-=20.5, T+=7.5
, (n=7)
T
Test statistic: T =
= 7.5
Conclusion: Dont reject the null hypothesis. No, the extra help session have
not been effective.
H0 :
rank
3
1
8
15
Tj
2
19
10
13
20
25
27
rank
9.5
2
4
12
16
43.5
3
16
20
26
19
rank
5.5
12
17
9.5
44
4
28
17
22
16
20
rank
18
7
14
5.5
12
56.5
H0 :
The two manuscripts are rated the same, p=0.5
H1 :
The weight losses for all four programs are the same.
H1 :
z z0.05 1.645
Rejection region:
Conclusion: Reject the null hypothesis. Yes, there is enough evidence to
conclude at the 5% significance level that manuscript 2 is more highly rated
than manuscript 1.
7.8147
Rejection region:
T j2
27 2 43.5 2 44 2 56.5 2
1683.15
4
5
4
5
n= 18
T j2
12
12
3 n 1
1683.15 3(19) 2.058
n(n 1)
nj
18(19)